GeoMotion: Rethinking Motion Segmentation via Latent 4D Geometry

2026-02-25

AI Summary

Motion segmentation in dynamic scenes is often hindered by noisy motion cues and the cumulative errors and high computational costs associated with multi-stage pipelines. This work proposes an end-to-end fully learned approach that, for the first time, integrates 4D geometric reconstruction priors—such as π³—with attention mechanisms to implicitly disentangle object and camera motion. By directly inferring moving objects from latent features, the method eliminates the need for explicit correspondence estimation or iterative optimization. Departing from conventional multi-stage frameworks, it introduces spatiotemporal priors into the learning process, achieving state-of-the-art motion segmentation performance across multiple benchmarks while significantly improving inference efficiency by circumventing complex preprocessing and optimization steps.

Technology Category

• Computer Vision: Motion & Tracking
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning
• Machine Learning: Learning with Manifolds

Application Category

• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation

Abstract

Motion segmentation in dynamic scenes is highly challenging, as conventional methods heavily rely on estimating camera poses and point correspondences from inherently noisy motion cues. Existing statistical inference or iterative optimization techniques that struggle to mitigate the cumulative errors in multi-stage pipelines often lead to limited performance or high computational cost. In contrast, we propose a fully learning-based approach that directly infers moving objects from latent feature representations via attention mechanisms, thus enabling end-to-end feed-forward motion segmentation. Our key insight is to bypass explicit correspondence estimation and instead let the model learn to implicitly disentangle object and camera motion. Supported by recent advances in 4D scene geometry reconstruction (e.g., $π^3$), the proposed method leverages reliable camera poses and rich spatial-temporal priors, which ensure stable training and robust inference for the model. Extensive experiments demonstrate that by eliminating complex pre-processing and iterative refinement, our approach achieves state-of-the-art motion segmentation performance with high efficiency. The code is available at:https://github.com/zjutcvg/GeoMotion.

Problem

• motion segmentation
• dynamic scenes
• camera pose estimation
• point correspondences
• cumulative errors

Innovation

• motion segmentation
• latent 4D geometry
• attention mechanism
• end-to-end learning
• camera pose priors